Integrated Genomic and Transcriptomic Analysis of Ductal

Human Cancer Biology

Integrated Genomic and Transcriptomic Analysis of Ductal Carcinoma In situ of the Breast Anne Vincent-Salomon,1, 2 Carlo Lucchesi,2 Nade' ge Gruel,2,3 Virginie Raynal,2 Gae«lle Pierron,1 Re¤mi Goudefroye,1Fabien Reyal,4 Franc¸ois Radvanyi,4 Re¤my Salmon,5 Jean-Paul Thiery,4 Xavier Sastre-Garau,1 Brigitte Sigal-Zafrani,1, 6 Alain Fourquet,7 and Olivier Delattre,1, 2 for the breast cancer study group of the Institut Curie

Abstract Purpose:To gain insight into genomic and trancriptomic subtypes of ductal carcinomas in situ of the breast (DCIS). Experimental Design: We did a combined phenotypic and genomic analysis of a series of 57 DCIS integrated with gene expression profile analysis for 26 of the 57 cases. Results: Thirty-two DCIS exhibited a luminal phenotype; 21 were ERBB2 positive, and 4 were ERBB2/estrogen receptor (ER) negative with1harboring a bona fide basal-like phenotype.Based on a CGH analysis, genomic types were identified in this series of DCIS with the 1qgain/16q loss combination observed in 3 luminal DCIS, the mixed amplifier pattern including all ERBB2,12lumi- nal and 2 ERBB2-/ER- DCIS, and the complex copy number alteration profile encompassing 14 luminal and 1ERBB2-/ER- DCIS. Eight cases (8 of 57; 14%) presented aTP53 mutation, all being amplifiers. Unsupervised analysis of gene expression profiles of 26 of the 57 DCIS showed that luminal and ERBB2-amplified, ER-negative cases clustered separately.We further investigated the effect of high and low copy number changes on gene expression. Strikingly, amplicons but also low copy number changes especially on 1q, 8q, and 16q in DCIS regulated the expression of a subset of genes in a very similar way to that recently described in invasive ductal carcinomas. Conclusions: These combined approaches show that the molecular heterogeneity of breast ductal carcinomas exists already in in situ lesions and further indicate that DCIS and invasive ductal carcinomas share genomic alterations with a similar effect on gene expression profile.

Ductal carcinoma in situ (DCIS) represents 15% to 20% of all They can be classified into three categories (low-, intermediate-, newly diagnosed cases of breast cancer, and its incidence is and high-grade) according to nuclear grade, differentiation, and increasing as a result of mammography screening (1). presence of central necrosis (2). However, this classification of DCIS correspond to a proliferation of malignant epithelial DCIS is poorly reproducible among pathologists (3). Molecular cells that spare the basal membrane with no stromal invasion. studies, still based on a limited number of cases, have revealed some characteristics of high-grade DCIS, including the high frequency of ERBB2 overexpression, TP53 mutations, chromo- Authors’ Affiliations: 1Institut Curie, Department of Tumor Biology, 2Institut somal 8p loss, and 1q gain (4–9). In contrast, positive estrogen National de la Sante et de la Recherche Medicale Unit 830, Institut Curie, 3Institut receptors (ER), BCL2 expression, and 16q loss have been Curie,Translational Research Department, 4Institut Curie, UMR144 Centre National reported to characterize low-grade DCIS (10–12). 5 6 de la Recherche Scientifique, Institut Curie, Department of Surgery, Head of the Treatment, regardless of DCIS nuclear grade, is based on mastec- Breast Cancer Study Group of the Institut Curie, and 7Institut Curie, Department of Radiation Oncology, Paris, France tomy or lumpectomy in combination with radiotherapy (1, 13). Received 6/14/07; revised 10/28/07; accepted 11/29/07. Factors known to be associated with a higher risk of recurrence Grant support: Institut National de la Sante et de la Recherche Medicale, the are young age, positive margins, necrosis, and high nuclear grade Institut Curie (Programme Incitatif et Coope¤ratif), and the Ligue Nationale contre le (14, 15). Some emerging data suggest that ERBB2-amplified Cancer. Construction of the BAC array was supported by grants from the Carte d’Identite¤des Tumeurs program of the Ligue Nationale Contre le Cancer. Dr Anne DCIS could present a higher risk of recurrence (16, 17). Vincent-Salomon was supported by an ‘‘Interface Institut National de la Sante et Recent gene expression profiling studies have reported that de la Recherche Medicale’’grant. consistent differences in expression of a subset of genes can be The costs of publication of this article were defrayed in part by the payment of page identified between low-grade and high-grade DCIS (18–20). charges. This article must therefore be hereby marked advertisement in accordance Interestingly, in situ lesions preferentially cluster with invasive with 18 U.S.C. Section 1734 solely to indicate this fact. Note: Supplementary data for this article are available at Clinical Cancer Research lesions of the same grade, suggesting that DCIS are precursors Online (http://clincancerres.aacrjournals.org/). of their similar grade invasive counterpart (20). In support of Requests for reprints: Olivier Delattre, Institut National de la Sante et this hypothesis, DCIS and invasive components of the same de la Recherche Medicale Unit 830, Institut Curie, 26 rue d’Ulm, 75248 Paris tumors exhibit similar patterns of genetic alterations (4, 7). Cedex 05, France. Phone: 33-1-42-34-66-81; Fax: 33-1-42-34-66-30; E-mail: [email protected]. Gene expression profile analysis classified invasive breast F 2008 American Association for Cancer Research. carcinomas into five groups: luminal A or B, basal like, ERBB2, doi:10.1158/1078-0432.CCR-07-1465 and normal like carcinomas (21, 22). These analyses have led

Clin Cancer Res 2008;14(7) April 1, 2008 1956 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2008 American Association for Cancer Research. Genetics of DCIS Subtypes of the Breast to the identification of specific immunophenotypic markers, Table 1. which allow this classification to be used in clinical practice Pathologic characteristics and (23, 24). It is well-recognized that ERBB2 is up-regulated more immunophenotype of the 57 DCIS frequently in DCIS than in invasive carcinomas (6). The identi- Variables All DCIS ERBB2 Luminal A ERBB2-/ER- fication of a basal-like entity is emerging with four recent studies describing DCIS lesions with this phenotype (18, 25–27). It n = 57* n = 21* n =32 n =4 (%) (%) (%) (%) nevertheless remains to be determined whether basal-like DCIS show specific genomic/transcriptomic alterations. Nuclear grade Genomic profiles also provide criteria for a clinically relevant High 29 (51) 17 (81) 9 (28) 3 (75) Nonhighc 28 (49) 4 (19) 23 (72) 1 (25) classification, as combined genomic and gene expression Necrosis studies in invasive cancers have highlighted the prognostic z10% 36 (63) 18 (86) 16 (50) 2 (50) effect of high level amplifications associated with gene over- <10% 21 (37) 3 (14) 16 (50) 2 (50) expression at chromosome regions 8p12, 11q13, 17q12, and ER 20q13 independently of the previously described major gene + 34 (60) 2 (10) 32 (100) 0 - 22 (40) 18 (90) 0 4 (100) expression classes. These cases are defined as amplifiers. Among PR nonamplifiers, one group with a good prognosis harbors few + 29 (52) 2 (10) 26 (81) 1 (25) genetic alterations but recurrent 16q loss associated with 1q + 27 (48) 18 (90) 6 (19) 3 (75) b gain (1q/16q). Another group, which includes most of basal- ERBB2 (IHC) like but also a subset of luminal cases, is characterized + 20 (36) 20 (100) 0 (0) 0 - 36 (64) 0 32 (100) 4 (100) by numerous low-level copy number alterations (CNA; KRT8/18 refs. 28, 29). In invasive breast cancer, this integrated genomic >50% 50 (89) 18 (90) 31 (97) 1 (25) approach, based on genomic and transcriptomic data, is very V50% 6 (11) 2 (10) 1 (3) 3 (75) promising to achieve better clinical management of patients. KRT5/6 z5% 15 (27) 5 (25) 9 (28) 1 (25) In the present study, we analyzed the immunophenotype, the <5% 41 (73) 15 (75) 23 (72) 3 (75) array CGH profiles, the TP53 gene sequence, and the gene EGFR expression profile of DCIS frozen samples. We identified DCIS z5% 6 (11) 3 (15) 2 (6) 1 (25) subtypes according to the molecular classification described for <5% 50 (89) 17 (85) 30 (94) 3 (75) invasive ductal carcinomas (IDC), allowing comparison of the genomic, transcriptomic, and phenotypic profiles of ERBB2, NOTE: Cases were considered positive for ER and PR using z10% luminal, and basal-like DCIS. Finally, correlations of genomic of positive nuclei per carcinomatous duct and for ERBB2 with a and transcriptomic data were used to identify candidate genes, cutoff of >60% of positive cells. Abbreviations: EGFR, epidermal growth factor receptor; KRT, the altered expression of which may play a critical role in early cytokeratin. breast carcinogenesis via copy number changes. *No fixed paraffin-embedded tissue sample was available for one case and immunophenotype was not assessed. Materials and Methods cNonhigh-grade tumors corresponded to 4 low-grade and 24 intermediate-grade DCIS. bIHC, ERBB2 status determined by immunohistochemistry. Selection of tumor samples. Fifty-seven DCIS of the breast were retrospectively selected from our tumor bank based on the availability of a frozen sample. Samples were collected on fresh surgical specimens by a pathologist within 1 hour after surgery from patients who TP53 sequencing. Exons 4 to 10 of TP53 were PCR amplified and underwent lumpectomy (n = 21) or mastectomy (n = 36), flash frozen bidirectionally sequenced using Big Dye Terminator chemistry (Applied in liquid nitrogen, and stored at -80jC. Experiments were done in Biosystems) with an ABI PRISM 3700 DNA Analyzer (primer sequences agreement with the French Bioethics Law 2004-800 and the Ethics previously published in ref. 32). Charter from the French National Institute of Cancer. All samples were Array CGH. The 3.5K BAC array together with experimental reviewed for histologic classification according to nuclear grade and procedures used for hybridization and washing were as previously classified as low, intermediate, and high nuclear grade (Table 1; ref. 2). published (32). All BAC and PAC were verified by end sequencing DNA and RNA extractions. All tumor samples contained >50% of before spotting. cancer cells, as assessed by H&E staining of histologic sections of the Analysis of array CGH data. Data analysis was based on the samples used for nucleic acid extraction. In this work, to avoid potential normalized ratios of Cy5/Cy3 signals observed for each BAC clone that artifacts, which may due to DNA or RNA amplification of microdissected previously passed the flag assessment procedure (32, 33). For samples, we privileged the investigation of samples large enough to autosomal chromosomes, the status of each BAC locus was defined extract good quality nucleic acids without need for preamplification according to the VAMP (Vizualization and Analysis of Molecular steps. DNA and RNA extractions were done using standard procedures as Profiles) analysis procedure (34) using the GLAD (Gain and Loss previously described (30). RNA quality control was checked on Agilent Analysis of DNA) algorithm that, based on the Adaptive Weights 2100 biolanalyzer (Agilent Technologies). Good quality RNA was Smoothing procedure, can be used to assign a status (gained, lost, or obtained in a subset of 26 cases among these 57 cases. normal) to each chromosome region (33). Immunohistochemistry. Immunostaining was done according to Regions of gain or loss and amplicons were defined as previously previously published protocols (31). The expression of ER (clone 6F11; published (32). BACs known to harbor copy number variation in 8 1/200; Novocastra), progesterone receptor (PR; clone 1A6; 1/200; public databases (Database of Genomic variants) or identified in the Novocastra), ERBB2 (clone CB11; 1/1,000; Novocastra), epidermal laboratory by array CGH experiments using normal/normal DNAs growth factor receptor (HER1; clone 31G7; 1/40; Zymed; Clinisciences), hybridizations were excluded from the analysis. cytokeratin 5/6 (clone D5/16B4; 1/50; Dako), and cytokeratin 8/18 (clone DC10; 1/100; Zymed; Clinisciences) were evaluated. For each anti- body, internal and external controls were included in the experiments. 8 http://projects.tcag.ca/variation/

www.aacrjournals.org 1957 Clin Cancer Res 2008;14(7) April1, 2008 Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2008 American Association for Cancer Research. Human Cancer Biology

Gene expression profile analysis. The DNA microarrays used in this radiation therapy (21 patients). The main tumor characteristics study were the Human Genome U133 Set (HG-U133; Affymetrix), are indicated in Table 1. Twenty-nine cases (51%) were high consisting of one GeneChip array (A) and containing almost 20,000 nuclear grade, and 28 (49%) were nonhigh-grade, including 24 probe sets. Microarray data were simultaneously normalized using the intermediate- and 4 low-grade tumors. For subsequent analy- GCRMA package version 1.1.4 in the R environment (R Development ses, the intermediate and low nuclear grade DCIS cases were Core Team). Unsupervised hierarchical clustering of the tumor sample was done using the dChip software (35) with standard Pearson merged into a nonhigh-grade group. The immunophenotype of correlation as similarity measure and centroid as linkage criteria. the 57 cases was determined to identify whether DCIS could be Clustering was done on a subset of probe sets with an interquartile classified into subgroups previously identified in IDC: luminal range of >2. Each probe set was preliminary standardized with respect ER-positive, ERBB2, basal-like triple-negative, and normal-like to mean and SD. The comparison of mean gene expression between carcinomas (Table 1). Altogether, these markers were able to ERBB2-amplified and luminal DCIS was done using SAM (36) and identify 32 luminal (ER positive) DCIS (most being nonhigh- Welch t tests. FDR for Welch t test was adjusted using the Benjamini- grade group; 23 cases), 21 ERBB2 DCIS (most being ER Hochberg corrections. Finally, for each probe set, the correlation with negative; 18 ER-negative cases), and 17 of 21 being high Cy5 to Cy3 ratios of the corresponding BAC clone was calculated using nuclear grade, and finally 4 cases were ER and ERBB2 negative. Pearson and Spearman correlation tests (for details see Supplementary Only one of these four ERBB2-negative/ER-negative DCIS was Data 1). PR negative, expressed cytokeratin 5/6 basal cytokeratins and epidermal growth factor receptor, and could therefore be Results considered to be a bona fide basal-like DCIS (Fig. 1A). Genomic analysis. We then investigated the TP53 status and Patient and tumor characteristics. Fifty-seven DCIS cases genomic profiles of these 57 DCIS lesions and their relation- with no associated invasive component were analyzed. The ship with the three main phenotypic classes identified above. median age of patients was 52.5 years (range, 30-79 years). All All samples were frozen DCIS lesions and comprised at least patients were treated according to established protocols: 50% of tumor cells. TP53 mutations were detected by direct mastectomy (36 patients) or conservative surgery followed by sequencing of exons 4 to 11. Mutations were found in eight

Fig. 1. Phenotypic and genomic analysis of DCIS. A, classification of the 57 DCIS according to the molecular subtypes previously identified for invasive breast carcinomas and their genomic status. *, three luminal DCIS and one ERBB2-negative, ER-negative DCIS presented a flat profile. B, examples of array CGH profiles from defined genomic status; B1, luminal with CNA; B2, luminal with 1qgain/16q loss profile; B3, luminal amplifier; B4, ERBB2 amplified; B5, triple-negative, basal-like amplifier with numerous CNA. C and D, frequencies of genome copy number gains and losses plotted as a function of genome location (C, luminal DCIS; D, ERBB2 -amplified DCIS). Orange, gains; green, losses. From chromosome 1pter on the left to chromosome Xq on the right.Vertical lines, chromosome boundaries; vertical hemi-dashed lines, centromere locations. The BAC clones exhibiting significantly more frequent gains (red) or losses (green)betweenluminal(C)andERBB2-amplified DCIS (D) are displayed underneath each frequency plot.

Clin Cancer Res 2008;14(7) April 1, 2008 1958 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2008 American Association for Cancer Research. Genetics of DCIS Subtypes of the Breast

Fig. 2. Unsupervised hierarchical clustering of 26 DCIS tumors using intrinsically variable gene expression of 502 genes (interquartile range, >2). Each row represents a different tumor, and each column represents one of 502 genes used for this unsupervised clustering (orange, ERBB2 amplified; yellow, ERBB2 not amplified), indicating the genomic type (red, amplifiers; gray, CNA; green, 1q/16q; white flat profile). Cyan, ER or PR positive; dark blue, ER and PR. NG, nuclear grade; black, high-grade; gray, nonhigh-grade.The genes are divided into two main clusters.The first cluster (1) contains genes involved in fatty acid metabolism, extracellular matrix, and the ERBB2 amplicon. The second cluster (2) includes genes involved in apoptosis and ER signaling.

cases (14%) and corresponded to two frameshift, two splice half of the amplicons (13 of 26) of the ERBB2 group associated site, two missense, and two stop mutations. These mutations with ERBB2 were also localized on chromosome 17q, but only were located in exon 5 (three cases); exon 6 (two cases); and 5 of 29 amplicons of the luminal group were localized on this exons 7, 8, and 9 (one case each). They occurred more chromosome arm (P = 2.10-2). frequently in high-grade DCIS (7 of 29; 24%) than in nonhigh- The ERBB2 amplicon size differed from one tumor to grade DCIS (1 of 28; 4%; P = 0.051) and significantly more another, ranging from 32.9 106 bp to 35.6 106 bp with a frequently in ERBB2 (6 of 17; 35%) than in luminal and minimal common region of amplification located between ERBB2-/ER- DCIS (2 of 36; 6%; P = 0.0047). 34.9 106 bp and 35.2 106 bp centered on the ERBB2 gene CNA (amplicons, gains, and losses) were identified using a sequence but also containing other genes including STARD3, one-megabase resolution BAC array. All samples but four GRB7, and PERLD1. exhibited altered profiles, indicating that the amount of stromal Regions of gains and losses observed with a frequency of or other normal cells in the samples was not sufficient to mask >20% in the luminal and ERBB2-amplified groups were copy number changes. Fifty-nine different amplicons were identified. Common regions of gain were observed in both observed in the 57 DCIS (Supplementary Data 2). Fourteen groups (Fig. 1C and D). They were located on chromosomes amplicons were recurrent (observed in at least two tumors), 1p, 1q, 5p, 7p, 7q, 8q, 14q, 16p, 19p, 19q, 17q, and 20q. Only including ERBB2 on chromosome 17q12, Cyclin D1 at 11q13, one common and recurrent region of loss on 11q was FGFR1 at 8p12, BMP7/BCAS1 at 20q13, but also five 17q identified. The ERBB2-amplified group presented specific amplicons distinct from ERBB2. Interestingly, the chromo- regions of 3p, 4p, 4q, and 8p losses and 17q gains compared somal localization of amplicons (distinct from ERBB2) differed with the luminal group that presented specific regions of gains between the ERBB2-amplified and the luminal DCIS groups, as on 1q, 8p, and 17q and losses on 16q (Supplementary Data 3).

www.aacrjournals.org 1959 Clin Cancer Res 2008;14(7) April1, 2008 Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2008 American Association for Cancer Research. Human Cancer Biology

According to the criteria recently proposed by Fridlyand and from the ERBB2 amplicon; the second was enriched in CNA Chin in 2006 (28, 29), DCIS cases could be classified into three (Fig. 2). categories: tumors characterized by the presence of amplicons Genomic and transcriptomic correlations. Given the strong (mixed amplifiers), tumors characterized by 1q gain and 16q relationship between the genomic classification and the loss (1q/16q group), and tumors with complex CNA. ERBB2- unsupervised clustering of gene expression profiles and to amplified DCIS never showed 16q loss. The luminal group determine the effect of genomic changes on gene expression exhibited genetic complexity with 12 cases classified as levels, we combined the analyses of the Affymetrix signals amplifiers, 3 cases harboring a 1q/16q profile, 14 cases and array CGH ratios for the 26 DCIS analyzed with both exhibiting a complex pattern of CNA, and 3 cases characterized approaches. Using the Pearson and Spearman correlation tests, by a flat profile. Within the ERBB2-/ER-negative DCIS group, we observed that the expression level of 416 genes was two belonged to the amplifier category with amplicons located significantly correlated with the ratio of their corresponding on chromosomes 5p, 8p, and 22q, one harbored CNA and one BAC locus (372 BAC clones; Fig. 3). Correlations corres- case was charade rised by a flat profile (Fig. 1A and B). ponding to amplicon were mostly located on 17q and Gene expression analysis. We also investigated transcrip- corresponding to either the variable extent of the ERBB2 tomic differences between ERBB2-amplified and luminal DCIS. amplicon from one tumor to another or to other 17q ampli- A subset of 26 cases for which good quality RNAs were cons, distinct from but co-occurring with ERBB2 (Supplemen- available, including 9 ERBB2-amplified and 17 luminal DCIS, tary Data 2). Other correlations due to amplicons mainly were analyzed on Affymetrix U133A. A total of 502 genes involved 8p (from 37 to 40 Mb), 11p12-p13 (from 34 to exhibited an interquartile range of >2 (Supplementary Data 4). 36 Mb), 11q13 (from 69 to 78 Mb), and 20q13.32 (from 46 to These genes discriminated two major clusters of tumors mainly 55 Mb; Supplementary Data 6). The main correlations due organized according to ERBB2 and hormonal receptor status to low copy gains or losses were observed on chromosomes rather than nuclear grade. One branch included most (8 of 9) 1q, 8q, and 16q (Table 2; Supplementary Data 6). The analysis ERBB2-amplified, ER-negative cases, and the other branch of genome expression correlations indicates that the amount comprised all (17 of 17) luminal cases and one ERBB2- of stromal or other normal cells in the samples was not amplified, ER-negative/PR-positive case. As shown in Fig. 2, sufficient to hamper the detection of a strong copy number genes coamplified with ERBB2 (GRB7, PERLD1, and STARD3) effect upon gene expression. and genes of the ER pathway (ESR1, GATA3, STC2, LIV1, Similar correlations have recently been described in IDC IGF1R...) were major contributors of this clustering. The most (28). Interestingly, when the lists of genes correlated with copy discriminant genes between ERBB2-amplified and luminal number changes in DCIS and IDC (28) were compared, we DCIS (Welch and Sam tests results, P < 0.01) are indicated in observed a strong overlap, as 204 of 416 (49%) genes that Supplementary Data 5. Unsupervised and supervised analyses correlated with genomic status in DCIS were also shown to be showed that genes involved in estrogens receptor pathways, correlated with genomic status in IDC. Strikingly, this overlap extracellular matrix remodeling (syndecan, ezrin, ...), immune represented 103 of 170 genes (61%) within the genomic response (TRBV19, ...), antiapoptotic pathways, cell adhesion regions differentially altered between the ERBB2 and luminal (CXCR4, syndecan, and ITGB6), or fatty acid metabolism DCIS groups (Table 2; Supplementary Data 3). (LDRL) were differentially expressed between the ERBB2- amplified and luminal groups (Fig. 2). Finally, genes encoding Discussion important signal transduction and transcription factors mole- cules were differentially expressed. FGFR2, RET, KIF5, and We report the first combined phenotypic, genomic, and IGF1R were overexpressed in luminal cases, whereas ERBB2, transcriptomic analysis of a series of DCIS frozen samples. The GRB7, PTPRC were overexpressed in ERBB2-amplified cases. most striking conclusions of this work are the high degree of Interestingly, displaying genomic characteristics of tumors on phenotypic, genomic, and transcriptomic similarities between the unsupervised cluster revealed that among the two branches DCIS and IDC and high diversity of breast tumors existing of the luminal cluster, one was enriched in amplifiers distinct already in situ early phase.

Fig. 3. Chromosomal localization of the most significant genome transcriptome correlations.The results of significant correlations according to Spearman and Pearson between probe sets and BAC loci are plotted as a function of genome location from chromosome1pter to chromosome Xq.

Clin Cancer Res 2008;14(7) April 1, 2008 1960 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2008 American Association for Cancer Research. Genetics of DCIS Subtypes of the Breast

Table 2. Genes showing significant expression copy number correlations and localized within regions discriminating ERBB2 and luminal DCIS

Chr Cytoband Startlocus Endlocus GeneID Correlation Correlation in IDC coefficient

1 q21.1 144150231 144356105 POLR3C 0.64 + 1 q21.1 144150231 144356105 PEX11B 0.63 + 1 q21.3 150870525 152158474 INTS3 0.68 1 q21.3 149888549 150060705 MRPL9 0.7 + 1 q23.1 154888337 155047911 MRPL24 0.76 + 1 q23.1 154888337 155047911 ISG20L2 0.63 + 1 q23.2 158452438 158743347 COPA 0.73 + 1 q23.2 158452438 158743347 NCSTN 0.66 + 1 q23.3 159044725 159381234 DEDD 0.75 + 1 q23.3 159514689 160713118 DUSP12 0.68 + 1 q23.3 159044725 159381234 PFDN2 0.66 + 1 q23.3 159381234 159514689 PPOX 0.76 + 1 q23.3 159381234 159514689 USP21 0.71 + 1 q23.3 159381234 159514689 B4GALT3 0.7 + 1 q23.3 159381234 159514689 NDUFS2 0.7 + 1 q24.1 165426500 165600956 POU2F1 0.62 1 q24.3 169943186 170838266 PIGC 0.77 + 1 q24.3 169943186 170838266 KIAA0859 0.71 + 1 q25.1 173402605 173547236 KIAA0040 0.63 + 1 q25.1 173189878 173350442 CACYBP 0.62 + 1 q25.3 180754428 181784986 LAMC2 0.74 1 q25.3 179011386 180565371 STX6 0.72 + 1 q32.1 204138581 204930767 IKBKE 0.82 1 q32.1 203342491 203495694 RIPK5 0.75 1 q32.1 202027575 202349129 ZC3H11A 0.76 1 q32.1 205087762 205277242 C1orf116 0.67 1 q32.1 203164295 203342491 RBBP5 0.75 1 q32.1 203963475 204138581 NUCKS1 0.71 1 q32.1 205277242 205431667 PFKFB2 0.67 1 q32.3 210152024 210772193 NENF 0.64 + 1 q41 219896304 221353763 MIA3 0.69 1 q42.11 221568617 223326349 NVL 0.74 + 1 q42.11 221568617 223326349 WDR26 0.74 + 1 q42.13 227197070 228112708 NUP133 0.72 + 1 q42.2 228271115 229195892 COG2 0.68 1 q42.3 232853266 233928437 RBM34 0.71 1 q42.3 232853266 233928437 TOMM20 0.72 1 q42.3 232853266 233928437 ARID4B 0.65 1 q43 234794031 243579104 HEATR1 0.83 + 8 p23.1 8827798 9852904 TNKS 0.82 + 8 p22 15734104 17364572 CNOT7 0.84 8 p21.3 22230663 22882560 BIN3 0.78 + 8 p21.2 24851358 26612249 PPP2R2A 0.81 + 8 p21.2 24851358 26612249 BNIP3L 0.68 + 8 p21.1 28844096 29353654 KIF13B 0.69 8 p12 36836067 38260917 GPR124 0.87 8 p12 36836067 38260917 ASH2L 0.81 + 8 p12 36836067 38260917 BRF2 0.8 + 8 p12 36836067 38260917 LSM1 0.8 + 8 p12 36836067 38260917 BAG4 0.79 + 8 p12 36836067 38260917 DDHD2 0.75 + 8 p12 29507083 31056165 DCTN6 0.77 8 p12 36836067 38260917 PROSC 0.69 + 8 p12 29507083 31056165 PPP2CB 0.64 8 p12 38335306 38425958 WHSC1L1 0.95 + 8 p12 38335306 38425958 FGFR1 0.92 + 8 p11.23 38497241 40010617 ADAM3A 0.79 8 p11.21 40185770 41825747 GINS4 0.79 8 p11.21 40185770 41825747 GOLGA7 0.73 8 p11.21 42044320 42283104 AP3M2 0.67 + 8 p11.21 42850482 43166289 FNTA 0.82 + 8 p11.21 42287374 42449080 VDAC3 0.78 + 8 p11.21 41825747 42044313 MYST3 0.77 + 8 q11.23 55029042 55683872 TCEA1 0.75 +

(Continued on the following page)

www.aacrjournals.org 1961 Clin Cancer Res 2008;14(7) April1, 2008 Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2008 American Association for Cancer Research. Human Cancer Biology

Table 2. Genes showing significant expression copy number correlations and localized within regions discriminating ERBB2 and luminal DCIS (Cont’d)

Chr Cytoband Startlocus Endlocus GeneID Correlation Correlation in IDC coefficient

8 q11.23 54844920 55029042 ATP6V1H 0.73 + 8 q12.1 58525613 59889959 LOC137886 0.74 8 q12.1 61401378 61814156 RAB2A 0.66 8 q21.11 74891351 76034750 TCEB1 0.67 8 q21.13 81648505 82728066 PMP2 0.67 8 q22.1 97062467 98198501 MTERFD1 0.81 + 8 q22.1 97062467 98198501 UQCRB 0.82 + 8 q22.1 97062467 98198501 PTDSS1 0.66 + 8 q22.1 95496349 95679222 RAD54B 0.77 + 8 q22.1 98198501 98375057 PGCP 0.74 8 q22.2 98375057 99735538 HRSP12 0.74 + 8 q22.2 100848701 102004873 POLR2K 0.74 + 8 q22.2 100738763 100848701 VPS13B 0.7 8 q22.3 103891151 104422933 ATP6V1C1 0.85 + 8 q22.3 103891151 104422933 AZIN1 0.7 + 8 q22.3 100848701 102004873 ANKRD46 0.7 8 q22.3 102195771 103646239 ZNF706 0.84 8 q23.2 110623681 110825964 EBAG9 0.76 + 8 q24.11 117853649 118046959 RAD21 0.67 + 8 q24.12 120401806 121248223 TAF2 0.73 + 8 q24.13 124908744 126133308 TRMT12 0.84 + 8 q24.13 124908744 126133308 RNF139 0.76 + 8 q24.13 123977707 124708761 DERL1 0.77 + 8 q24.13 124908744 126133308 SQLE 0.64 + 8 q24.13 123977707 124708761 C8orf32 0.65 + 8 q24.13 126133308 126334773 NSMCE2 0.74 8 q24.22 131964334 133156759 KIAA0143 0.8 + 8 q24.3 143909082 145536611 SIAHBP1 0.85 + 8 q24.3 140031123 141603135 NIBP 0.83 8 q24.3 143909082 145536611 EXOSC4 0.82 + 8 q24.3 143909082 145536611 CYC1 0.8 + 8 q24.3 143909082 145536611 GPAA1 0.73 + 8 q24.3 143909082 145536611 SHARPIN 0.72 8 q24.3 146133391 C8orf33 0.72 + 8 q24.3 143909082 145536611 PYCRL 0.72 + 8 q24.3 143909082 145536611 GRINA 0.71 8 q24.3 143909082 145536611 SCRIB 0.68 + 8 q24.3 143909082 145536611 PLEC1 0.68 8 q24.3 143909082 145536611 ZNF696 0.68 + 8 q24.3 143909082 145536611 ZC3H3 0.66 + 8 q24.3 141603135 141808565 PTK2 0.79 + 8 q24.3 146029133 146133391 ZNF7 0.71 + 8 q24.3 143739886 143909082 C8orf55 0.69 8 q24.3 146029133 146133391 ZNF16 0.66 + 16 q12.1 46027393 46605539 PHKB 0.69 + 16 q12.1 48832865 49022383 BRD7 0.74 16 q13 56003087 56156112 COQ9 0.75 16 q13 56003087 56156112 CIAPIN1 0.71 16 q13 54836075 56003087 ARL2BP 0.68 16 q21 56619483 57476831 CNOT1 0.83 + 16 q21 56619483 57476831 GOT2 0.71 + 16 q21 65020904 65209402 CKLF 0.64 16 q22.1 65209402 65623027 APPBP1 0.75 16 q22.1 65209402 65623027 FAM96B 0.65 + 16 q22.1 67478745 69245806 AARS 0.72 + 16 q22.1 67478745 69245806 DDX19A 0.72 + 16 q22.1 67478745 69245806 SF3B3 0.72 + 16 q22.1 67478745 69245806 WWP2 0.67 + 16 q22.1 69245806 69419474 VAC14 0.7 + 16 q22.3 69419474 71533759 KIAA0174 0.85 + 16 q22.3 69419474 71533759 DHX38 0.71 + 16 q22.3 71533759 74776431 PSMD7 0.8 + 16 q22.3 71533759 74776431 RFWD3 0.72 + 16 q23.1 71533759 74776431 KARS 0.79 +

(Continued on the following page)

Clin Cancer Res 2008;14(7) April 1, 2008 1962 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2008 American Association for Cancer Research. Genetics of DCIS Subtypes of the Breast

Table 2. Genes showing significant expression copy number correlations and localized within regions discriminating ERBB2 and luminal DCIS (Cont’d)

Chr Cytoband Startlocus Endlocus GeneID Correlation Correlation in IDC coefficient

16 q23.1 71533759 74776431 ADAT1 0.73 16 q23.1 71533759 74776431 WDR59 0.68 16 q23.1 71533759 74776431 TERF2IP 0.64 + 16 q23.2 79052806 80639519 ASCIZ 0.72 + 16 q24.1 83335109 84444005 COX4NB 0.82 + 16 q24.1 83335109 84444005 KIAA0182 0.71 16 q24.1 83335109 84444005 COX4I1 0.69 + 16 q24.1 83335109 84444005 ZDHHC7 0.69 + 17 p13.3 2077107 3789432 ITGAE 0.8 17 q11.2 24125209 25739938 ERAL1 0.66 17 q11.2 24125209 25739938 FLOT2 0.75 17 q11.2 22465093 23921490 IFT20 0.71 17 q11.2 23991111 24125209 RPL23A 0.77 17 q11.2 23926892 24057159 SUPT6H 0.76 17 q11.2 27772588 27941694 PSMD11 0.76 + 17 q11.2 23991111 24125209 C17orf63 0.76 17 q11.2 23944011 24039852 SPAG5 0.75 17 q11.2 23926892 24057159 SDF2 0.68 17 q11.2 26430455 26574472 NF1 0.72 17 q11.2 23991111 24125209 TRAF4 0.63 17 q12 34434812 34987922 CRKRS 0.82 17 q12 34434812 34987922 RPL19 0.85 17 q12 34434812 34987922 PPARBP 0.71 17 q12 34987922 35196413 GRB7 0.87 + 17 q12 35246223 35393977 GSDML 0.87 + 17 q12 34987922 35196413 PERLD1 0.86 + 17 q12 34987922 35196413 STARD3 0.82 + 17 q12 34987922 35196413 ERBB2 0.82 + 17 q12 35246223 35393977 PSMD3 0.81 + 17 q12 30384409 30599689 NLE1 0.76 + 17 q12 34201212 34324738 CCDC49 0.75 17 q12 30270215 30599689 LIG3 0.73 + 17 q12 34201212 34324738 RPL23 0.71 17 q12 34201212 34324738 PIP5K2B 0.68 17 q12 34987922 35196413 PNMT 0.65 + 17 q21.1 35465293 35659097 CASC3 0.66 + 17 q21.1 35465293 35659097 THRA 0.66 17 q21.2 35920386 37443438 SMARCE1 0.66 + 17 q21.2 35682074 35839525 CDC6 0.81 17 q21.2 35682074 35839525 WIPF2 0.75 + 17 q21.2 35737736 35915957 RARA 0.71 17 q21.2 35682074 35839525 TOP2A 0.7

NOTE: The genes are ordered by chromosomes and are located in regions differentially gained or lost between ERBB2 and luminal DCIS (listed in Supplementary Data 3). The correlation coefficients (Pearson and Spearman) were considered significant when higher than 0.6 with a Benjamini-Hochberg corrected P value lower than 0.05. In case of multiple probe sets corresponding to one gene, the highest correlation coefficient is indicated. Similarly, for probe sets that matched with two flanking (backward and forward) BACs, the highest correlation coefficient is shown. The last column indicates the genes exhibiting expression copy number correlations in infiltrating ductal carcinomas (28). The complete list of genes exhibiting significant expression copy number correlations is provided in Supplementary Data 6.

DCIS can be classified into the luminal, ERBB2, and basal- ing that the recently recognized basal-like DCIS group is a rare like groups previously described in IDC (21, 22). In our series entity among pure DCIS (6-10% of cases; refs. 25, 26, 38, 39). of 57 DCIS, most tumors (56%) belonged to the luminal group This was also shown by a recent gene expression profile analysis based on ER expression and the absence of ERBB2 over- of a series of DCIS (18). expression. The second group of tumors comprised ERBB2 When compared with the luminal group, the ERBB2- overexpressing DCIS (37%) that were mostly high grade (81%) amplified cases presented a particular array CGH profile with and ER negative/PR negative (90%) in agreement with previous more frequent 17q gains, 3p, 4p, 4q, and 8p losses. In contrast, reports showing that ERBB2 overexpression is rare in ER- the luminal DCIS group harbored significantly more frequent positive DCIS (37). A third small group was composed of three 16q losses. These observations refine previously published triple-negative DCIS (5%) with only one case demonstrating a classic CGH data showing a significantly higher number of 17q clear basal-like immunophenotype (2%) with positive staining gains in high nuclear grade DCIS (which correspond to most for keratin 5/6 and epidermal growth factor receptor, confirm- cases of our ERBB2-amplified, ER-negative group) and 16q

www.aacrjournals.org 1963 Clin Cancer Res 2008;14(7) April1, 2008 Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2008 American Association for Cancer Research. Human Cancer Biology losses in low-grade DCIS (which are represented in the present Strikingly, a large part of the genes, for which a strong series by the luminal cases; refs. 4, 7, 11). correlation between expression level and genomic status was We also show that DCIS can be classified according to the observed, also exhibited a similar correlation in IDC (28), genomic criteria recently proposed for IDC (28), as a first group indicating that genome alterations have similar consequences of 34 tumors presented high-level amplification of genomic on gene expression in both IDC and DCIS. In line with the regions (amplifiers). All TP53 mutations were observed in this data reported by Chin et al. (28), these results identify genes group. Remarkably, the recurrent amplicons of DCIS were located in frequently altered regions in breast cancer, strongly located on 8p12 (FGFR1), 11q12 (Cyclin D1), 17q12 (ERBB2), influenced by copy number variations, and whose gene and 20q13 (BMP7/BCAS1), corresponding also to the most fre- dosage effects may play a role in early phases of breast quent amplicons of IDC (28). Compared with other amplifier carcinogenesis. Two genes on chromosome 16 seem to be of cases and as recently observed in IDC (28), half amplicons particular interest: ASCIZ (ATM/ATR-substrate ChK2-interacting associated with ERBB2 were also located on chromosome 17q. Zinc finger; ref. 52), which encodes a protein involved in In contrast, in the luminal group, only 17% of amplicons lesion-specific Rad51 focus formation, a process that also involved 17q. A second group comprised tumors with a 16q involves BRCA1 and BRCA2, two master genes in breast deletion associated with 1q gain but no additional CNA. All carcinogenesis; and CNOT1, which encodes a member of the these cases presented a luminal phenotype. The last group CCR4-NOT complex, a global regulator of transcription that comprised cases with more complex CNA. In this series, the may act as a repressor of endogenous estrogens target genes three triple-negative DCIS did not present any recurrent geno- (53). Interestingly, CNOT7, another member of the CCR4- mic profile, but two were amplifiers including the bona fide NOT complex, is regulated by gene copy number variation basal-like DCIS, which also harbored a TP53 gene mutation. on chromosome 8p, a chromosome frequently altered in Unsupervised analysis of gene expression profiles supports ERBB2 DCIS tumors. Based on these observation, the the observation that ERBB2-amplified and luminal DCIS hypothesis of haploinsufficiency of the CCR4-NOT complex constitute two distinct entities. Gene ontology analysis of the playing a role in early breast tumorigenesis may now be ERBB2 DCIS signature showed that the genes up-regulated in tested. this group belonged to ERBB2 signal transduction pathways Altogether, these integrated genomic and transcriptomic data but also to fatty acid metabolism and other protein kinase confirm previous results showing strong similarities between and transmembrane signaling pathways. Genes down-regulat- IDC and DCIS (20, 54). Although we cannot exclude the ed mainly belonged to apoptotic and ER regulation path- possibility that subtle genetic alterations that may have escaped ways. The list of genes differentially expressed between detection in this study could also play a role in the evolution ERBB2 and luminal DCIS in this series was compared with from DCIS to IDC, this study raises the hypothesis that the the recently published list of genes that discriminate DCIS difference between DCIS and IDC is not based on specific according to their differentiation status (18). Although the genetic alterations or on their consequences on gene expression two comparisons were based on slightly different criteria, profile. Because a recent report suggested that a 35-gene they both highlighted major genes involved in breast expression classifier might distinguish IDC and DCIS (18), carcinogenesis such as ESR1, GRB7, and ERBB2. A link the role of epigenetic modifications in the progression from between the ERBB2 pathway and fatty acid metabolism has DCIS to IDC will need to be evaluated. been previously reported in cell lines (40) and invasive breast In conclusion, our data show that DCIS already displays the carcinomas (41). The present study also showed that these molecular diversity observed in IDC and, therefore, can be pathways are activated right from the earliest stages of classified according to molecular criteria distinguishing ERBB2- ERBB2-driven breast carcinogenesis. Interestingly, as reported amplified DCIS, usually high-grade, ER negative with frequent in IDC (42, 43), overexpression of 17q12 genes STARD3, TP53 mutations, from luminal DCIS corresponding to low/ GRB7, and PERLD1 is also associated with ERBB2 over- intermediate-grade, ER positive with a very low rate of TP53 expression in DCIS, which raises the question of their mutations. In our series, only three cases were classified as triple putative synergistic role during early breast carcinogenesis. negative and only one was classified as basal-like DCIS, which Finally, Integrin b6, a receptor for fibronectin and cytotactin, is confirms that this last entity is rare among DCIS. Further overexpressed in the ERBB2 DCIS group. Recent data have studies are needed to define whether a classification of DCIS highlighted molecular interactions between ERBB2 receptors based on molecular markers may help to more accurately and integrin networks (41, 44). Moreover, experiments con- define cases associated with a higher risk of recurrence. And ducted with ITGB6 knockout mice indicate that ITGB6 may finally, genomic/transcriptomic correlations represent a promi- regulate transforming growth factor h and metalloprotease, in sing tool to identify new genes and pathways important in early particular matrix metalloproteinase 12 expression (45). Further breast carcinogenesis. investigations are needed to understand the role of ITGB6 in ERBB2-amplified DCIS. It is noteworthy that genes recently involved in angiogenesis Acknowledgments or invasion such as SDC1 (syndecan; refs. 46–48) or CXCR4 are We thank Dr Paul Fre¤neaux for his constant support during our study, Ce¤line up-regulated in ERBB2-positive in situ lesions (49). Reciprocally, Grassart for her excellent technical assistance, and Ingrid Lebigot and her team from FGFR2, a gene recently linked to in susceptibility locus in breast the Institut Curie’s tumor bank. cancer (50, 51), is up-regulated in luminal DCIS. Members of the Breast Cancer Study group in addition to coauthors: Bernard Asselain, Alain Aurias, Emmanuel Barillot, Marc Bollet, Franc¸ois Campana, Paul The correlation between array CGH and expression data Cottu, Patricia de Cremoux, Ve¤ronique Die¤ras, Laurent Mignot, Jean-Yves Pierga, shows that amplifications but also more subtle copy number Marie-France Poupon, Dominique Stoppa-Lyonnet, Anne Tardivon, Fabienne changes lead to modifications of the gene expression profile. Thibault, and PascaleThis.

Clin Cancer Res 2008;14(7) April1,2008 1964 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2008 American Association for Cancer Research. Genetics of DCIS Subtypes of the Breast

References 1. Burstein HJ, Polyak K,WongJS, et al. Ductal carcino- 18. HannemannJ,Velds A, HalfwerkJB, et al. Classifica- 37. Collins LC, Schnitt SJ. HER2 protein overexpression ma in situ of the breast. N Engl J Med 2004;350: tion of ductal carcinoma in situ by gene expression in estrogen receptor-positive ductal carcinoma in situ 1430^41. profiling. Breast Cancer Res 2006;8:R61. of the breast: frequency and implications for tamoxi- 2. Recht A, Rutgers EJ, Fentiman IS, et al. The fourth 19. Adeyinka A, Emberley E, Niu Y, et al. Analysis of fen therapy. Mod Pathol 2005;18:615^20. eortc DCIS Consensus Meeting (Chateau Marquette, gene expression in ductal carcinoma in situ of the 38. Livasy CA, Perou CM, Karaca G, et al. Identification Heemskerk, the Netherlands, 23-24 January 1998)- breast. Clin Cancer Res 2002;8:3788 ^95. of a basal-like subtype of breast ductal carcinoma Conference Report. Eur J Cancer 1998;34:1664 ^ 9. 20. Ma X-J, Salunga R,Tuggle JT, et al. Gene expression in situ. Hum Pathol 2007;38:197 ^ 204. 3. Sneige N, Lagios MD, Schwarting R, et al. Interob- profiles of human breast cancer progression. Proc Natl 39. Paredes J, Lopes N, Milanezi F, et al. P-cadherin server reproducibility of the Lagios nuclear grading Acad Sci U S A 2003;100:5974^ 9. and cytokeratin 5: useful adjunct markers to distin- system for ductal carcinoma in situ. Hum Pathol 1999; 21. Perou CM, Sorlie T, Eisen MB, et al. Molecular por- guish basal-like ductal carcinomas in situ.Virchows 30:257 ^ 62. traits of human breast tumours. Nature 2000;406: Arch 2007;450:73^80. 4. O’Connell P,Pekkel W, Fuqua S, et al. Analysis of loss 747 ^ 52. 40. Kumar-Sinha C, Ignatoski KW, Lippman ME, et al. of heterozygosity in 399 premalignant breast lesions at 22. SorlieT, Perou CM,Tibshirani R, et al. Gene expres- Transcriptome analysis of HER2 reveals a molecular 15geneticloci. JNatl Cancer Inst1998;90:697 ^ 703. sion patterns of breast carcinomas distinguish tumor connection to fatty acid synthesis. Cancer Res 2003; 5. Simpson JF, Page DL. The p53 tumor suppressor subclasses with clinical implications. Proc Natl Acad 63:132^9. gene in ductal carcinoma in situ of the breast. Am J Sci U S A 2001;98:10869 ^ 74. 41. Bertucci F, Borie N, Ginestier C, et al. Identification Pathol 2000;156:5 ^ 6. 23. Nielsen TO, Hsu FD, Jensen K, et al. Immunohisto- and validation of an ERBB2 gene expression signature 6. Van de Vijver MJ, Peterse JL, Mooi WJ, et al. Neu- chemical and clinical characterization of the basal-like in breast cancers. Oncogene 2004;23:2564 ^ 75. protein overexpression in breast cancer. Association subtype of invasive breast carcinoma. Clin Cancer Res 42. Kauraniemi P, Kallioniemi A. Activation of multi- with comedo-type ductal carcinoma in situ and limit- 2004;10:5367 ^ 74. ple cancer-associated genes at the ERBB2 ampli- ed prognostic value in stage II breast cancer. N Engl J 24. van de Rijn M, Perou CM,Tibshirani R, et al. Expres- con in breast cancer. Endocr Relat Cancer 2006; Med 1988;319:1239 ^ 45. sion of cytokeratins 17 and 5 identifies a group of 13 :3 9 ^ 4 9. 7. Hwang ES, DeVries S, Chew KL, et al. Patterns of breast carcinomas with poor clinical outcome. Am J 43. Kauraniemi P, Kuukasjarvi T, Sauter G, et al. Amplifi- chromosomal alterations in breast ductal carcinoma Pathol 2002;161:1991 ^ 6. cation of a 280-kilobase core region at the ERBB2 in situ. Clin Cancer Res 2004;10:5160 ^ 7. 25. BryanBB,SchnittSJ,CollinsLC.Ductalcarcinoma locus leads to activation of two hypothetical proteins 8. Done S, Eskandarian S, Bull S, et al. p53 missense in situ with basal-like phenotype: a possible precursor in breast cancer. Am J Pathol 2003;163:1979 ^ 84. mutations in microdissected high-grade ductal carci- to invasive basal-like breast cancer. Mod Pathol 2006; 44. Guo W, PylayevaY, Pepe A, et al. h4integrinampli- noma in situ of the breast. J Natl Cancer Inst 2001; 19 :617 ^ 21. fies ERBB2 signaling to promote mammary tumori- 93:700 ^ 4. 26. Dabbs DJ, Chivukula M, Carter G, et al. Basal genesis. Cell 2006;126:489 ^ 502. 9. Buerger H, Mommers E, Littmann R, et al. Correlation phenotype of ductal carcinoma in situ: recognition 45. Morris DG, Huang X, Kaminski N, et al. Loss of of morphologic and cytogenetic parameters of genetic and immunohistologic profile. Mod Pathol 2006;19: integrin a(v)h6-mediated TGF-h activation causes instability with chromosomal alterations in in situ car- 1506^11. Mmp12-dependent emphysema. Nature 2003;422: cinomas of the breast. Am J Clin Pathol 2000;114: 27. Livasy CA, Karaca G, Nanda R, et al. Phenotypic 169 ^ 73. 854^9. evaluation of the basal-like subtype of invasive breast 46. MaedaT, Desouky J, Friedl A. Syndecan-1 expres- 10. Zafrani B, Leroyer A, Fourquet A, et al. Mammo- carcinoma. Mod Pathol 2006;19:264 ^ 71. sion by stromal fibroblasts promotes breast carcinoma graphically-detected ductal in situ carcinoma of the 28. Chin K, DeVries S, Fridlyand J, et al. Genomic and growth in vivo and stimulates tumor angiogenesis. breast analyzed with a new classification. A study of transcriptional aberrations linked to breast cancer Oncogene 2006;25:1408 ^ 12. 127 cases: correlation with estrogen and progester- pathophysiologies. Cancer Cell 2006;10:529 ^ 41. 47. Baba F, Swartz K, van Buren R, et al. Syndecan-1 one receptors, p53 and c-erbb-2 proteins, and prolif- 29. FridlyandJ, Snijders AM,Ylstra B, et al. Breast tumor and syndecan-4 are overexpressed in an estrogen re- erative activity. Semin Diagn Pathol 1994;11:208 ^14. copy number aberration phenotypes and genomic in- ceptor-negative, highly proliferative breast carcinoma 11. Buerger H, Otterbach F, Simon R, et al. Comparative stability. BMC Cancer 2006;6:96. subtype. Breast Cancer ResTreat 2006;98:91 ^ 8. genomic hybridization of ductal carcinoma in situ of 30. Reyal F, Stransky N, Bernard-Pierrot I, et al. Visual- 48. Gotte M, Kersting C, Radke I, et al. An expression the breast-evidence of multiple genetic pathways. izing chromosomes as transcriptome correlation signature of syndecan-1 (cd138), E-cadherin and JPathol1999;187:396^402. maps: evidence of chromosomal domains containing c-met is associated with factors of angiogenesis and 12. Quinn CM, Ostrowski JL, Harkins L, et al. Loss of co-expressed genes-a study of 130 invasive ductal lymphangiogenesis in ductal breast carcinoma in situ. bcl-2 expression in ductal carcinoma in situ of the breast carcinomas. Cancer Res 2005;65:1376 ^ 83. Breast Cancer Res 2007;9:R8. breast relates to poor histological differentiation and 31. Azoulay S, Lae M, Freneaux P, et al. KIT is highly 49. Li YM, PanY,Wei Y, et al. Upregulation of CXCR4 is to expression of p53 and c-erbb-2 proteins [in Pro- expressed in adenoid cystic carcinoma of the breast, essential for HER2-mediated tumor metastasis. Can- cess Citation]. Histopathology 1998;33:531 ^ 6. a basal-like carcinoma associated with a favorable cer Cell 2004;6:459 ^ 69. 13. Leonard GD, Swain SM. Ductal carcinoma in situ, outcome. Mod Pathol 2005;18:1623 ^ 31. 50. Easton DF, Pooley KA, Dunning AM, et al. Ge- complexities and challenges. J Natl Cancer Inst 2004; 32.Vincent-Salomon A, Gruel N, Lucchesi C, et al. Iden- nome-wide association study identifies novel breast 96:906^20. tification of typical medullary breast carcinoma as a cancer susceptibility loci. Nature 2007;447:1087 ^ 93. 14. Sanders ME, Schuyler PA, Dupont WD, et al. The genomic sub-group of basal-like carcinomas, a het- 51. Hunter DJ, Kraft P,Jacobs KB, et al. A genome-wide natural history of low-grade ductal carcinoma in situ erogeneous new molecular entity. Breast Cancer Res association study identifies alleles in FGFR2 associat- of the breast in women treated by biopsy only 2007;9:R24. ed with risk of sporadic postmenopausal breast can- revealed over 30 years of long-term follow-up. Cancer 33. Hupe P, Stransky N,ThieryJP, et al. Analysis of array cer. Nat Genet 2007;39:870 ^ 4. 2005;103:2481 ^ 4. CGH data: from signal ratio to gain and loss of dna 52. McNees CJ, Conlan LA,Tenis N, et al. ASCIZ regu- 15. Wong JS, Kaelin CM, Troyan SL, et al. Prospective regions. Bioinformatics 2004;20:3413^ 22. lates lesion-specific rad51 focus formation and apo- study of wide excision alone for ductal carcinoma 34. La Rosa P,Viara E, Hupe P,et al.VAMP: visualization ptosis after methylating DNA damage. EMBO J in situ of the breast. J Clin Oncol 2006;24:1031 ^ 6. and analysis of array-CGH, transcriptome and other 2005;24:2447^57. 16. Kepple J, Henry-Tillman RS, Klimberg VS, et al. The molecular profiles. Bioinformatics 2006;22:2066 ^ 73. 53. Winkler GS, Mulder KW, Bardwell VJ, et al. Human receptor expression pattern in ductal carcinoma in situ 35. Li C, Wang WH. DNA-Chip Analyzer (dChip). In: Ccr4-Not complex is a ligand-dependent repressor of predicts recurrence. Am J Surg 2006;192:68 ^71. ‘‘The Analysis of Gene Expression Data: Methods nuclear receptor-mediated transcription. EMBO J 17. Provenzano E, HopperJL, Giles GG, et al. Biological and Software’’. NewYork: Springer. 2006;25:3089 ^ 99. markers that predict clinical recurrence in ductal carci- 36.Tusher VG,Tibshirani R, Chu G. Significance analysis 54. Porter D, Lahti-Domenici J, Keshaviah A, et al. noma in situ of the breast. Eur J Cancer 2003;39: of microarrays applied to the ionizing radiation re- Molecular markers in ductal carcinoma in situ of the 622 ^ 30. sponse. Proc Natl Acad Sci U S A 2001;98:5116^ 21. breast. Mol Cancer Res 2003;1:362 ^ 75.

www.aacrjournals.org 1965 Clin Cancer Res 2008;14(7) April1, 2008 Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2008 American Association for Cancer Research. Integrated Genomic and Transcriptomic Analysis of Ductal Carcinoma In situ of the Breast

Anne Vincent-Salomon, Carlo Lucchesi, Nadège Gruel, et al.

Clin Cancer Res 2008;14:1956-1965.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/14/7/1956

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2008/03/31/14.7.1956.DC1

Cited articles This article cites 53 articles, 11 of which you can access for free at: http://clincancerres.aacrjournals.org/content/14/7/1956.full#ref-list-1

Citing articles This article has been cited by 14 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/14/7/1956.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/14/7/1956. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.